character.java

一个开源的JAVA虚拟机

	    if (c <= 0x20FF)
		return COMBINING_MARKS_FOR_SYMBOLS;
	    if (c <= 0x214F)
		return LETTERLIKE_SYMBOLS;
	    if (c <= 0x218F)
		return NUMBER_FORMS;
	    if (c <= 0x21FF)
		return ARROWS;
	    if (c <= 0x22FF)
		return MATHEMATICAL_OPERATORS;
	    if (c <= 0x23FF)
		return MISCELLANEOUS_TECHNICAL;
	    if (c <= 0x243F)
		return CONTROL_PICTURES;
	    if (c <= 0x245F)
		return OPTICAL_CHARACTER_RECOGNITION;
	    if (c <= 0x24FF)
		return ENCLOSED_ALPHANUMERICS;
	    if (c <= 0x257F)
		return BOX_DRAWING;
	    if (c <= 0x259F)
		return BLOCK_ELEMENTS;
	    if (c <= 0x25FF)
		return GEOMETRIC_SHAPES;
	    if (c <= 0x26FF)
		return MISCELLANEOUS_SYMBOLS;
	    if (c <= 0x27BF)
		return DINGBATS;
	    if (c <= 0x2FFF)
		return null;
	    if (c <= 0x303F)
		return CJK_SYMBOLS_AND_PUNCTUATION;
	    if (c <= 0x309F)
		return HIRAGANA;
	    if (c <= 0x30FF)
		return KATAKANA;
	    if (c <= 0x312F)
		return BOPOMOFO;
	    if (c <= 0x318F)
		return HANGUL_COMPATIBILITY_JAMO;
	    if (c <= 0x319F)
		return KANBUN;
	    if (c <= 0x31FF)
		return null;
	    if (c <= 0x32FF)
		return ENCLOSED_CJK_LETTERS_AND_MONTHS;
	    if (c <= 0x33FF)
		return CJK_COMPATIBILITY;
	    if (c <= 0x4DFF)
		return null;
	    if (c <= 0x9FFF)
		return CJK_UNIFIED_IDEOGRAPHS;
	    if (c <= 0xABFF)
		return null;
	    if (c <= 0xD7A3)
		return HANGUL_SYLLABLES;
	    if (c <= 0xD7FF)
		return null;
	    if (c <= 0xDFFF)
		return SURROGATES_AREA;
	    if (c <= 0xF8FF)
		return PRIVATE_USE_AREA;
	    if (c <= 0xFAFF)
		return CJK_COMPATIBILITY_IDEOGRAPHS;
	    if (c <= 0xFB4F)
		return ALPHABETIC_PRESENTATION_FORMS;
	    if (c <= 0xFDFF)
		return ARABIC_PRESENTATION_FORMS_A;
	    if (c <= 0xFE1F)
		return null;
	    if (c <= 0xFE2F)
		return COMBINING_HALF_MARKS;
	    if (c <= 0xFE4F)
		return CJK_COMPATIBILITY_FORMS;
	    if (c <= 0xFE6F)
		return SMALL_FORM_VARIANTS;
	    if (c <= 0xFEFE)
		return ARABIC_PRESENTATION_FORMS_B;
	    if (c <= 0xFEFF)
		return SPECIALS;
	    if (c <= 0xFFEF)
		return HALFWIDTH_AND_FULLWIDTH_FORMS;
	    return SPECIALS;
	}
    }

    public static final int MIN_RADIX = 2;
    public static final int MAX_RADIX = 36;
    public static final char MIN_VALUE = '\u0000';
    public static final char MAX_VALUE = '\uffff';
    public static final Class TYPE = Class.getPrimitiveType(Class.T_CHAR);

    public static final byte UNASSIGNED             = 0;
    public static final byte UPPERCASE_LETTER       = 1;
    public static final byte LOWERCASE_LETTER       = 2;
    public static final byte TITLECASE_LETTER       = 3;
    public static final byte MODIFIER_LETTER        = 4;
    public static final byte OTHER_LETTER           = 5;
    public static final byte NON_SPACING_MARK       = 6;
    public static final byte ENCLOSING_MARK         = 7;
    public static final byte COMBINING_SPACING_MARK = 8;
    public static final byte DECIMAL_DIGIT_NUMBER   = 9;
    public static final byte LETTER_NUMBER          = 10;
    public static final byte OTHER_NUMBER           = 11;
    public static final byte SPACE_SEPARATOR        = 12;
    public static final byte LINE_SEPARATOR         = 13;
    public static final byte PARAGRAPH_SEPARATOR    = 14;
    public static final byte CONTROL                = 15;
    public static final byte FORMAT                 = 16;
    public static final byte PRIVATE_USE            = 18;
    public static final byte SURROGATE              = 19;
    public static final byte DASH_PUNCTUATION       = 20;
    public static final byte START_PUNCTUATION      = 21;
    public static final byte END_PUNCTUATION        = 22;
    public static final byte CONNECTOR_PUNCTUATION  = 23;
    public static final byte OTHER_PUNCTUATION      = 24;
    public static final byte MATH_SYMBOL            = 25;
    public static final byte CURRENCY_SYMBOL        = 26;
    public static final byte MODIFIER_SYMBOL        = 27;
    public static final byte OTHER_SYMBOL           = 28;

    private static final byte Mn = NON_SPACING_MARK;
    private static final byte Mc = COMBINING_SPACING_MARK;
    private static final byte Me = ENCLOSING_MARK;
    private static final byte Nd = DECIMAL_DIGIT_NUMBER;
    private static final byte Nl = LETTER_NUMBER;
    private static final byte No = OTHER_NUMBER;
    private static final byte Zs = SPACE_SEPARATOR;
    private static final byte Zl = LINE_SEPARATOR;
    private static final byte Zp = PARAGRAPH_SEPARATOR;
    private static final byte Cc = CONTROL;
    private static final byte Cf = FORMAT;
    private static final byte Cs = SURROGATE;
    private static final byte Co = PRIVATE_USE;
    private static final byte Cn = UNASSIGNED;
    private static final byte Lu = UPPERCASE_LETTER;
    private static final byte Ll = LOWERCASE_LETTER;
    private static final byte Lt = TITLECASE_LETTER;
    private static final byte Lm = MODIFIER_LETTER;
    private static final byte Lo = OTHER_LETTER;
    private static final byte Pc = CONNECTOR_PUNCTUATION;
    private static final byte Pd = DASH_PUNCTUATION;
    private static final byte Ps = START_PUNCTUATION;
    private static final byte Pe = END_PUNCTUATION;
    private static final byte Po = OTHER_PUNCTUATION;
    private static final byte Sm = MATH_SYMBOL;
    private static final byte Sc = CURRENCY_SYMBOL;
    private static final byte Sk = MODIFIER_SYMBOL;
    private static final byte So = OTHER_SYMBOL;

    private char value;

    public Character(char value) {
	this.value = value;
    }

    public char charValue() {
	return value;
    }
    public int hashCode() {
	return (int) value;
    }
    public boolean equals(Object obj) {
	return
	    obj instanceof Character &&
	    value == ((Character) obj).value;
    }
    public String toString() {
	return String.valueOf(value);
    }
    public static boolean isLowerCase(char ch) {
	return getType(ch) == Ll;
    }
    public static boolean isUpperCase(char ch) {
	return getType(ch) == Lu;
    }
    public static boolean isTitleCase(char ch) {
	return getType(ch) == Lt;
    }
    public static boolean isDigit(char ch) {
	return getType(ch) == Nd;
    }
    public static boolean isDefined(char ch) {
	return
	    getType(ch) != Cn ||
	    (0x3040 <= ch && ch <= 0x9FA5) ||
	    (0xF900 <= ch && ch <= 0xFA2D);
    }
    public static boolean isLetter(char ch) {
	switch (getType(ch)) {
	case Lu:
	case Ll:
	case Lt:
	case Lm:
	case Lo:
	    return true;
	default:
	    return false;
	}
    }
    public static boolean isLetterOrDigit(char ch) {
	switch (getType(ch)) {
	case Lu:
	case Ll:
	case Lt:
	case Lm:
	case Lo:
	case Nd:
	    return true;
	default:
	    return false;
	}
    }
    public static boolean isJavaLetter(char ch) {
	return isLetter(ch) || ch == '$' || ch == '_';
    }
    public static boolean isJavaLetterOrDigit(char ch) {
	return isLetterOrDigit(ch) || ch == '$' || ch == '_';
    }
    public static boolean isJavaIdentifierStart(char ch) {
	switch (getType(ch)) {
	case Lu:
	case Ll:
	case Lt:
	case Lm:
	case Lo:
	case Sc:
	case Pc:
	    return true;
	default:
	    return false;
	}
    }
    public static boolean isJavaIdentifierPart(char ch) {
	switch (getType(ch)) {
	case Lu:
	case Ll:
	case Lt:
	case Lm:
	case Lo:
	case Nd:
	case Sc:
	case Pc:
	case Nl:
	case Mc:
	case Mn:
	    return true;
	default:
	    return isIdentifierIgnorable(ch);
	}
    }
    public static boolean isUnicodeIdentifierStart(char ch) {
	switch (getType(ch)) {
	case Lu:
	case Ll:
	case Lt:
	case Lm:
	case Lo:
	    return true;